vislang's picture
Upload folder using huggingface_hub
84514b1 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.6867673104495235,
"eval_steps": 500,
"global_step": 5000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 5e-05,
"loss": 0.7302,
"step": 10
},
{
"epoch": 0.01,
"learning_rate": 5e-05,
"loss": 0.6767,
"step": 20
},
{
"epoch": 0.01,
"learning_rate": 5e-05,
"loss": 0.6744,
"step": 30
},
{
"epoch": 0.01,
"learning_rate": 5e-05,
"loss": 0.6699,
"step": 40
},
{
"epoch": 0.02,
"learning_rate": 5e-05,
"loss": 0.6799,
"step": 50
},
{
"epoch": 0.02,
"learning_rate": 5e-05,
"loss": 0.6763,
"step": 60
},
{
"epoch": 0.02,
"learning_rate": 5e-05,
"loss": 0.674,
"step": 70
},
{
"epoch": 0.03,
"learning_rate": 5e-05,
"loss": 0.6733,
"step": 80
},
{
"epoch": 0.03,
"learning_rate": 5e-05,
"loss": 0.668,
"step": 90
},
{
"epoch": 0.03,
"learning_rate": 5e-05,
"loss": 0.6679,
"step": 100
},
{
"epoch": 0.04,
"learning_rate": 5e-05,
"loss": 0.6722,
"step": 110
},
{
"epoch": 0.04,
"learning_rate": 5e-05,
"loss": 0.6707,
"step": 120
},
{
"epoch": 0.04,
"learning_rate": 5e-05,
"loss": 0.6737,
"step": 130
},
{
"epoch": 0.05,
"learning_rate": 5e-05,
"loss": 0.6697,
"step": 140
},
{
"epoch": 0.05,
"learning_rate": 5e-05,
"loss": 0.6663,
"step": 150
},
{
"epoch": 0.05,
"learning_rate": 5e-05,
"loss": 0.6793,
"step": 160
},
{
"epoch": 0.06,
"learning_rate": 5e-05,
"loss": 0.6706,
"step": 170
},
{
"epoch": 0.06,
"learning_rate": 5e-05,
"loss": 0.6718,
"step": 180
},
{
"epoch": 0.06,
"learning_rate": 5e-05,
"loss": 0.6725,
"step": 190
},
{
"epoch": 0.07,
"learning_rate": 5e-05,
"loss": 0.6749,
"step": 200
},
{
"epoch": 0.07,
"learning_rate": 5e-05,
"loss": 0.6748,
"step": 210
},
{
"epoch": 0.07,
"learning_rate": 5e-05,
"loss": 0.6719,
"step": 220
},
{
"epoch": 0.08,
"learning_rate": 5e-05,
"loss": 0.6722,
"step": 230
},
{
"epoch": 0.08,
"learning_rate": 5e-05,
"loss": 0.6671,
"step": 240
},
{
"epoch": 0.08,
"learning_rate": 5e-05,
"loss": 0.6688,
"step": 250
},
{
"epoch": 0.09,
"learning_rate": 5e-05,
"loss": 0.6701,
"step": 260
},
{
"epoch": 0.09,
"learning_rate": 5e-05,
"loss": 0.6828,
"step": 270
},
{
"epoch": 0.09,
"learning_rate": 5e-05,
"loss": 0.6716,
"step": 280
},
{
"epoch": 0.1,
"learning_rate": 5e-05,
"loss": 0.6725,
"step": 290
},
{
"epoch": 0.1,
"learning_rate": 5e-05,
"loss": 0.6706,
"step": 300
},
{
"epoch": 0.1,
"learning_rate": 5e-05,
"loss": 0.6669,
"step": 310
},
{
"epoch": 0.11,
"learning_rate": 5e-05,
"loss": 0.667,
"step": 320
},
{
"epoch": 0.11,
"learning_rate": 5e-05,
"loss": 0.6725,
"step": 330
},
{
"epoch": 0.11,
"learning_rate": 5e-05,
"loss": 0.6635,
"step": 340
},
{
"epoch": 0.12,
"learning_rate": 5e-05,
"loss": 0.6661,
"step": 350
},
{
"epoch": 0.12,
"learning_rate": 5e-05,
"loss": 0.6624,
"step": 360
},
{
"epoch": 0.12,
"learning_rate": 5e-05,
"loss": 0.6657,
"step": 370
},
{
"epoch": 0.13,
"learning_rate": 5e-05,
"loss": 0.664,
"step": 380
},
{
"epoch": 0.13,
"learning_rate": 5e-05,
"loss": 0.6605,
"step": 390
},
{
"epoch": 0.13,
"learning_rate": 5e-05,
"loss": 0.6635,
"step": 400
},
{
"epoch": 0.14,
"learning_rate": 5e-05,
"loss": 0.6664,
"step": 410
},
{
"epoch": 0.14,
"learning_rate": 5e-05,
"loss": 0.6696,
"step": 420
},
{
"epoch": 0.15,
"learning_rate": 5e-05,
"loss": 0.658,
"step": 430
},
{
"epoch": 0.15,
"learning_rate": 5e-05,
"loss": 0.6649,
"step": 440
},
{
"epoch": 0.15,
"learning_rate": 5e-05,
"loss": 0.6611,
"step": 450
},
{
"epoch": 0.16,
"learning_rate": 5e-05,
"loss": 0.6575,
"step": 460
},
{
"epoch": 0.16,
"learning_rate": 5e-05,
"loss": 0.6703,
"step": 470
},
{
"epoch": 0.16,
"learning_rate": 5e-05,
"loss": 0.665,
"step": 480
},
{
"epoch": 0.17,
"learning_rate": 5e-05,
"loss": 0.6617,
"step": 490
},
{
"epoch": 0.17,
"learning_rate": 5e-05,
"loss": 0.6623,
"step": 500
},
{
"epoch": 0.17,
"learning_rate": 5e-05,
"loss": 0.6557,
"step": 510
},
{
"epoch": 0.18,
"learning_rate": 5e-05,
"loss": 0.6558,
"step": 520
},
{
"epoch": 0.18,
"learning_rate": 5e-05,
"loss": 0.663,
"step": 530
},
{
"epoch": 0.18,
"learning_rate": 5e-05,
"loss": 0.6627,
"step": 540
},
{
"epoch": 0.19,
"learning_rate": 5e-05,
"loss": 0.6528,
"step": 550
},
{
"epoch": 0.19,
"learning_rate": 5e-05,
"loss": 0.6586,
"step": 560
},
{
"epoch": 0.19,
"learning_rate": 5e-05,
"loss": 0.659,
"step": 570
},
{
"epoch": 0.2,
"learning_rate": 5e-05,
"loss": 0.6583,
"step": 580
},
{
"epoch": 0.2,
"learning_rate": 5e-05,
"loss": 0.6594,
"step": 590
},
{
"epoch": 0.2,
"learning_rate": 5e-05,
"loss": 0.6454,
"step": 600
},
{
"epoch": 0.21,
"learning_rate": 5e-05,
"loss": 0.6572,
"step": 610
},
{
"epoch": 0.21,
"learning_rate": 5e-05,
"loss": 0.6546,
"step": 620
},
{
"epoch": 0.21,
"learning_rate": 5e-05,
"loss": 0.656,
"step": 630
},
{
"epoch": 0.22,
"learning_rate": 5e-05,
"loss": 0.6544,
"step": 640
},
{
"epoch": 0.22,
"learning_rate": 5e-05,
"loss": 0.6574,
"step": 650
},
{
"epoch": 0.22,
"learning_rate": 5e-05,
"loss": 0.6528,
"step": 660
},
{
"epoch": 0.23,
"learning_rate": 5e-05,
"loss": 0.6574,
"step": 670
},
{
"epoch": 0.23,
"learning_rate": 5e-05,
"loss": 0.6566,
"step": 680
},
{
"epoch": 0.23,
"learning_rate": 5e-05,
"loss": 0.6512,
"step": 690
},
{
"epoch": 0.24,
"learning_rate": 5e-05,
"loss": 0.6517,
"step": 700
},
{
"epoch": 0.24,
"learning_rate": 5e-05,
"loss": 0.6461,
"step": 710
},
{
"epoch": 0.24,
"learning_rate": 5e-05,
"loss": 0.6562,
"step": 720
},
{
"epoch": 0.25,
"learning_rate": 5e-05,
"loss": 0.6616,
"step": 730
},
{
"epoch": 0.25,
"learning_rate": 5e-05,
"loss": 0.6606,
"step": 740
},
{
"epoch": 0.25,
"learning_rate": 5e-05,
"loss": 0.6463,
"step": 750
},
{
"epoch": 0.26,
"learning_rate": 5e-05,
"loss": 0.6494,
"step": 760
},
{
"epoch": 0.26,
"learning_rate": 5e-05,
"loss": 0.6549,
"step": 770
},
{
"epoch": 0.26,
"learning_rate": 5e-05,
"loss": 0.644,
"step": 780
},
{
"epoch": 0.27,
"learning_rate": 5e-05,
"loss": 0.6535,
"step": 790
},
{
"epoch": 0.27,
"learning_rate": 5e-05,
"loss": 0.6489,
"step": 800
},
{
"epoch": 0.27,
"learning_rate": 5e-05,
"loss": 0.6477,
"step": 810
},
{
"epoch": 0.28,
"learning_rate": 5e-05,
"loss": 0.6389,
"step": 820
},
{
"epoch": 0.28,
"learning_rate": 5e-05,
"loss": 0.6459,
"step": 830
},
{
"epoch": 0.28,
"learning_rate": 5e-05,
"loss": 0.6444,
"step": 840
},
{
"epoch": 0.29,
"learning_rate": 5e-05,
"loss": 0.648,
"step": 850
},
{
"epoch": 0.29,
"learning_rate": 5e-05,
"loss": 0.6353,
"step": 860
},
{
"epoch": 0.29,
"learning_rate": 5e-05,
"loss": 0.6719,
"step": 870
},
{
"epoch": 0.3,
"learning_rate": 5e-05,
"loss": 0.6566,
"step": 880
},
{
"epoch": 0.3,
"learning_rate": 5e-05,
"loss": 0.64,
"step": 890
},
{
"epoch": 0.3,
"learning_rate": 5e-05,
"loss": 0.6476,
"step": 900
},
{
"epoch": 0.31,
"learning_rate": 5e-05,
"loss": 0.6469,
"step": 910
},
{
"epoch": 0.31,
"learning_rate": 5e-05,
"loss": 0.6385,
"step": 920
},
{
"epoch": 0.31,
"learning_rate": 5e-05,
"loss": 0.649,
"step": 930
},
{
"epoch": 0.32,
"learning_rate": 5e-05,
"loss": 0.6551,
"step": 940
},
{
"epoch": 0.32,
"learning_rate": 5e-05,
"loss": 0.6371,
"step": 950
},
{
"epoch": 0.32,
"learning_rate": 5e-05,
"loss": 0.6512,
"step": 960
},
{
"epoch": 0.33,
"learning_rate": 5e-05,
"loss": 0.6486,
"step": 970
},
{
"epoch": 0.33,
"learning_rate": 5e-05,
"loss": 0.6461,
"step": 980
},
{
"epoch": 0.33,
"learning_rate": 5e-05,
"loss": 0.6617,
"step": 990
},
{
"epoch": 0.34,
"learning_rate": 5e-05,
"loss": 0.654,
"step": 1000
},
{
"epoch": 0.34,
"learning_rate": 5e-05,
"loss": 0.6395,
"step": 1010
},
{
"epoch": 0.34,
"learning_rate": 5e-05,
"loss": 0.6582,
"step": 1020
},
{
"epoch": 0.35,
"learning_rate": 5e-05,
"loss": 0.6447,
"step": 1030
},
{
"epoch": 0.35,
"learning_rate": 5e-05,
"loss": 0.6378,
"step": 1040
},
{
"epoch": 0.35,
"learning_rate": 5e-05,
"loss": 0.6377,
"step": 1050
},
{
"epoch": 0.36,
"learning_rate": 5e-05,
"loss": 0.633,
"step": 1060
},
{
"epoch": 0.36,
"learning_rate": 5e-05,
"loss": 0.639,
"step": 1070
},
{
"epoch": 0.36,
"learning_rate": 5e-05,
"loss": 0.6396,
"step": 1080
},
{
"epoch": 0.37,
"learning_rate": 5e-05,
"loss": 0.6491,
"step": 1090
},
{
"epoch": 0.37,
"learning_rate": 5e-05,
"loss": 0.6396,
"step": 1100
},
{
"epoch": 0.37,
"learning_rate": 5e-05,
"loss": 0.6542,
"step": 1110
},
{
"epoch": 0.38,
"learning_rate": 5e-05,
"loss": 0.6399,
"step": 1120
},
{
"epoch": 0.38,
"learning_rate": 5e-05,
"loss": 0.648,
"step": 1130
},
{
"epoch": 0.38,
"learning_rate": 5e-05,
"loss": 0.6614,
"step": 1140
},
{
"epoch": 0.39,
"learning_rate": 5e-05,
"loss": 0.6453,
"step": 1150
},
{
"epoch": 0.39,
"learning_rate": 5e-05,
"loss": 0.6441,
"step": 1160
},
{
"epoch": 0.39,
"learning_rate": 5e-05,
"loss": 0.6381,
"step": 1170
},
{
"epoch": 0.4,
"learning_rate": 5e-05,
"loss": 0.6449,
"step": 1180
},
{
"epoch": 0.4,
"learning_rate": 5e-05,
"loss": 0.6443,
"step": 1190
},
{
"epoch": 0.4,
"learning_rate": 5e-05,
"loss": 0.6411,
"step": 1200
},
{
"epoch": 0.41,
"learning_rate": 5e-05,
"loss": 0.6371,
"step": 1210
},
{
"epoch": 0.41,
"learning_rate": 5e-05,
"loss": 0.6329,
"step": 1220
},
{
"epoch": 0.41,
"learning_rate": 5e-05,
"loss": 0.642,
"step": 1230
},
{
"epoch": 0.42,
"learning_rate": 5e-05,
"loss": 0.6364,
"step": 1240
},
{
"epoch": 0.42,
"learning_rate": 5e-05,
"loss": 0.6349,
"step": 1250
},
{
"epoch": 0.43,
"learning_rate": 5e-05,
"loss": 0.6341,
"step": 1260
},
{
"epoch": 0.43,
"learning_rate": 5e-05,
"loss": 0.65,
"step": 1270
},
{
"epoch": 0.43,
"learning_rate": 5e-05,
"loss": 0.645,
"step": 1280
},
{
"epoch": 0.44,
"learning_rate": 5e-05,
"loss": 0.6401,
"step": 1290
},
{
"epoch": 0.44,
"learning_rate": 5e-05,
"loss": 0.6329,
"step": 1300
},
{
"epoch": 0.44,
"learning_rate": 5e-05,
"loss": 0.6373,
"step": 1310
},
{
"epoch": 0.45,
"learning_rate": 5e-05,
"loss": 0.6379,
"step": 1320
},
{
"epoch": 0.45,
"learning_rate": 5e-05,
"loss": 0.6383,
"step": 1330
},
{
"epoch": 0.45,
"learning_rate": 5e-05,
"loss": 0.6452,
"step": 1340
},
{
"epoch": 0.46,
"learning_rate": 5e-05,
"loss": 0.6312,
"step": 1350
},
{
"epoch": 0.46,
"learning_rate": 5e-05,
"loss": 0.6334,
"step": 1360
},
{
"epoch": 0.46,
"learning_rate": 5e-05,
"loss": 0.6395,
"step": 1370
},
{
"epoch": 0.47,
"learning_rate": 5e-05,
"loss": 0.6425,
"step": 1380
},
{
"epoch": 0.47,
"learning_rate": 5e-05,
"loss": 0.6422,
"step": 1390
},
{
"epoch": 0.47,
"learning_rate": 5e-05,
"loss": 0.6393,
"step": 1400
},
{
"epoch": 0.48,
"learning_rate": 5e-05,
"loss": 0.6357,
"step": 1410
},
{
"epoch": 0.48,
"learning_rate": 5e-05,
"loss": 0.6274,
"step": 1420
},
{
"epoch": 0.48,
"learning_rate": 5e-05,
"loss": 0.6421,
"step": 1430
},
{
"epoch": 0.49,
"learning_rate": 5e-05,
"loss": 0.6319,
"step": 1440
},
{
"epoch": 0.49,
"learning_rate": 5e-05,
"loss": 0.6441,
"step": 1450
},
{
"epoch": 0.49,
"learning_rate": 5e-05,
"loss": 0.6327,
"step": 1460
},
{
"epoch": 0.5,
"learning_rate": 5e-05,
"loss": 0.632,
"step": 1470
},
{
"epoch": 0.5,
"learning_rate": 5e-05,
"loss": 0.6351,
"step": 1480
},
{
"epoch": 0.5,
"learning_rate": 5e-05,
"loss": 0.6187,
"step": 1490
},
{
"epoch": 0.51,
"learning_rate": 5e-05,
"loss": 0.6307,
"step": 1500
},
{
"epoch": 0.51,
"learning_rate": 5e-05,
"loss": 0.6434,
"step": 1510
},
{
"epoch": 0.51,
"learning_rate": 5e-05,
"loss": 0.6367,
"step": 1520
},
{
"epoch": 0.52,
"learning_rate": 5e-05,
"loss": 0.6391,
"step": 1530
},
{
"epoch": 0.52,
"learning_rate": 5e-05,
"loss": 0.6417,
"step": 1540
},
{
"epoch": 0.52,
"learning_rate": 5e-05,
"loss": 0.6305,
"step": 1550
},
{
"epoch": 0.53,
"learning_rate": 5e-05,
"loss": 0.6391,
"step": 1560
},
{
"epoch": 0.53,
"learning_rate": 5e-05,
"loss": 0.6404,
"step": 1570
},
{
"epoch": 0.53,
"learning_rate": 5e-05,
"loss": 0.635,
"step": 1580
},
{
"epoch": 0.54,
"learning_rate": 5e-05,
"loss": 0.6339,
"step": 1590
},
{
"epoch": 0.54,
"learning_rate": 5e-05,
"loss": 0.6352,
"step": 1600
},
{
"epoch": 0.54,
"learning_rate": 5e-05,
"loss": 0.6346,
"step": 1610
},
{
"epoch": 0.55,
"learning_rate": 5e-05,
"loss": 0.6261,
"step": 1620
},
{
"epoch": 0.55,
"learning_rate": 5e-05,
"loss": 0.6372,
"step": 1630
},
{
"epoch": 0.55,
"learning_rate": 5e-05,
"loss": 0.632,
"step": 1640
},
{
"epoch": 0.56,
"learning_rate": 5e-05,
"loss": 0.6318,
"step": 1650
},
{
"epoch": 0.56,
"learning_rate": 5e-05,
"loss": 0.6439,
"step": 1660
},
{
"epoch": 0.56,
"learning_rate": 5e-05,
"loss": 0.646,
"step": 1670
},
{
"epoch": 0.57,
"learning_rate": 5e-05,
"loss": 0.6389,
"step": 1680
},
{
"epoch": 0.57,
"learning_rate": 5e-05,
"loss": 0.6267,
"step": 1690
},
{
"epoch": 0.57,
"learning_rate": 5e-05,
"loss": 0.6339,
"step": 1700
},
{
"epoch": 0.58,
"learning_rate": 5e-05,
"loss": 0.6305,
"step": 1710
},
{
"epoch": 0.58,
"learning_rate": 5e-05,
"loss": 0.6241,
"step": 1720
},
{
"epoch": 0.58,
"learning_rate": 5e-05,
"loss": 0.6325,
"step": 1730
},
{
"epoch": 0.59,
"learning_rate": 5e-05,
"loss": 0.6367,
"step": 1740
},
{
"epoch": 0.59,
"learning_rate": 5e-05,
"loss": 0.6333,
"step": 1750
},
{
"epoch": 0.59,
"learning_rate": 5e-05,
"loss": 0.6284,
"step": 1760
},
{
"epoch": 0.6,
"learning_rate": 5e-05,
"loss": 0.627,
"step": 1770
},
{
"epoch": 0.6,
"learning_rate": 5e-05,
"loss": 0.6327,
"step": 1780
},
{
"epoch": 0.6,
"learning_rate": 5e-05,
"loss": 0.6326,
"step": 1790
},
{
"epoch": 0.61,
"learning_rate": 5e-05,
"loss": 0.6384,
"step": 1800
},
{
"epoch": 0.61,
"learning_rate": 5e-05,
"loss": 0.6752,
"step": 1810
},
{
"epoch": 0.61,
"learning_rate": 5e-05,
"loss": 0.6355,
"step": 1820
},
{
"epoch": 0.62,
"learning_rate": 5e-05,
"loss": 0.6387,
"step": 1830
},
{
"epoch": 0.62,
"learning_rate": 5e-05,
"loss": 0.6319,
"step": 1840
},
{
"epoch": 0.62,
"learning_rate": 5e-05,
"loss": 0.63,
"step": 1850
},
{
"epoch": 0.63,
"learning_rate": 5e-05,
"loss": 0.6361,
"step": 1860
},
{
"epoch": 0.63,
"learning_rate": 5e-05,
"loss": 0.6288,
"step": 1870
},
{
"epoch": 0.63,
"learning_rate": 5e-05,
"loss": 0.6311,
"step": 1880
},
{
"epoch": 0.64,
"learning_rate": 5e-05,
"loss": 0.6319,
"step": 1890
},
{
"epoch": 0.64,
"learning_rate": 5e-05,
"loss": 0.6394,
"step": 1900
},
{
"epoch": 0.64,
"learning_rate": 5e-05,
"loss": 0.6236,
"step": 1910
},
{
"epoch": 0.65,
"learning_rate": 5e-05,
"loss": 0.6308,
"step": 1920
},
{
"epoch": 0.65,
"learning_rate": 5e-05,
"loss": 0.6302,
"step": 1930
},
{
"epoch": 0.65,
"learning_rate": 5e-05,
"loss": 0.6333,
"step": 1940
},
{
"epoch": 0.66,
"learning_rate": 5e-05,
"loss": 0.6223,
"step": 1950
},
{
"epoch": 0.66,
"learning_rate": 5e-05,
"loss": 0.6379,
"step": 1960
},
{
"epoch": 0.66,
"learning_rate": 5e-05,
"loss": 0.6307,
"step": 1970
},
{
"epoch": 0.67,
"learning_rate": 5e-05,
"loss": 0.6268,
"step": 1980
},
{
"epoch": 0.67,
"learning_rate": 5e-05,
"loss": 0.6312,
"step": 1990
},
{
"epoch": 0.67,
"learning_rate": 5e-05,
"loss": 0.6396,
"step": 2000
},
{
"epoch": 0.68,
"learning_rate": 5e-05,
"loss": 0.6417,
"step": 2010
},
{
"epoch": 0.68,
"learning_rate": 5e-05,
"loss": 0.6363,
"step": 2020
},
{
"epoch": 0.68,
"learning_rate": 5e-05,
"loss": 0.6305,
"step": 2030
},
{
"epoch": 0.69,
"learning_rate": 5e-05,
"loss": 0.6294,
"step": 2040
},
{
"epoch": 0.69,
"learning_rate": 5e-05,
"loss": 0.6301,
"step": 2050
},
{
"epoch": 0.69,
"learning_rate": 5e-05,
"loss": 0.6286,
"step": 2060
},
{
"epoch": 0.7,
"learning_rate": 5e-05,
"loss": 0.6282,
"step": 2070
},
{
"epoch": 0.7,
"learning_rate": 5e-05,
"loss": 0.6296,
"step": 2080
},
{
"epoch": 0.71,
"learning_rate": 5e-05,
"loss": 0.6372,
"step": 2090
},
{
"epoch": 0.71,
"learning_rate": 5e-05,
"loss": 0.627,
"step": 2100
},
{
"epoch": 0.71,
"learning_rate": 5e-05,
"loss": 0.6477,
"step": 2110
},
{
"epoch": 0.72,
"learning_rate": 5e-05,
"loss": 0.6352,
"step": 2120
},
{
"epoch": 0.72,
"learning_rate": 5e-05,
"loss": 0.6346,
"step": 2130
},
{
"epoch": 0.72,
"learning_rate": 5e-05,
"loss": 0.6284,
"step": 2140
},
{
"epoch": 0.73,
"learning_rate": 5e-05,
"loss": 0.6261,
"step": 2150
},
{
"epoch": 0.73,
"learning_rate": 5e-05,
"loss": 0.6646,
"step": 2160
},
{
"epoch": 0.73,
"learning_rate": 5e-05,
"loss": 0.6357,
"step": 2170
},
{
"epoch": 0.74,
"learning_rate": 5e-05,
"loss": 0.6307,
"step": 2180
},
{
"epoch": 0.74,
"learning_rate": 5e-05,
"loss": 0.6334,
"step": 2190
},
{
"epoch": 0.74,
"learning_rate": 5e-05,
"loss": 0.6287,
"step": 2200
},
{
"epoch": 0.75,
"learning_rate": 5e-05,
"loss": 0.6252,
"step": 2210
},
{
"epoch": 0.75,
"learning_rate": 5e-05,
"loss": 0.6286,
"step": 2220
},
{
"epoch": 0.75,
"learning_rate": 5e-05,
"loss": 0.6332,
"step": 2230
},
{
"epoch": 0.76,
"learning_rate": 5e-05,
"loss": 0.6312,
"step": 2240
},
{
"epoch": 0.76,
"learning_rate": 5e-05,
"loss": 0.6397,
"step": 2250
},
{
"epoch": 0.76,
"learning_rate": 5e-05,
"loss": 0.6234,
"step": 2260
},
{
"epoch": 0.77,
"learning_rate": 5e-05,
"loss": 0.6236,
"step": 2270
},
{
"epoch": 0.77,
"learning_rate": 5e-05,
"loss": 0.6274,
"step": 2280
},
{
"epoch": 0.77,
"learning_rate": 5e-05,
"loss": 0.6281,
"step": 2290
},
{
"epoch": 0.78,
"learning_rate": 5e-05,
"loss": 0.6376,
"step": 2300
},
{
"epoch": 0.78,
"learning_rate": 5e-05,
"loss": 0.6397,
"step": 2310
},
{
"epoch": 0.78,
"learning_rate": 5e-05,
"loss": 0.6317,
"step": 2320
},
{
"epoch": 0.79,
"learning_rate": 5e-05,
"loss": 0.6173,
"step": 2330
},
{
"epoch": 0.79,
"learning_rate": 5e-05,
"loss": 0.6272,
"step": 2340
},
{
"epoch": 0.79,
"learning_rate": 5e-05,
"loss": 0.6336,
"step": 2350
},
{
"epoch": 0.8,
"learning_rate": 5e-05,
"loss": 0.6273,
"step": 2360
},
{
"epoch": 0.8,
"learning_rate": 5e-05,
"loss": 0.6328,
"step": 2370
},
{
"epoch": 0.8,
"learning_rate": 5e-05,
"loss": 0.6336,
"step": 2380
},
{
"epoch": 0.81,
"learning_rate": 5e-05,
"loss": 0.6269,
"step": 2390
},
{
"epoch": 0.81,
"learning_rate": 5e-05,
"loss": 0.6265,
"step": 2400
},
{
"epoch": 0.81,
"learning_rate": 5e-05,
"loss": 0.6357,
"step": 2410
},
{
"epoch": 0.82,
"learning_rate": 5e-05,
"loss": 0.6264,
"step": 2420
},
{
"epoch": 0.82,
"learning_rate": 5e-05,
"loss": 0.6228,
"step": 2430
},
{
"epoch": 0.82,
"learning_rate": 5e-05,
"loss": 0.6228,
"step": 2440
},
{
"epoch": 0.83,
"learning_rate": 5e-05,
"loss": 0.6203,
"step": 2450
},
{
"epoch": 0.83,
"learning_rate": 5e-05,
"loss": 0.6184,
"step": 2460
},
{
"epoch": 0.83,
"learning_rate": 5e-05,
"loss": 0.6266,
"step": 2470
},
{
"epoch": 0.84,
"learning_rate": 5e-05,
"loss": 0.6244,
"step": 2480
},
{
"epoch": 0.84,
"learning_rate": 5e-05,
"loss": 0.6258,
"step": 2490
},
{
"epoch": 0.84,
"learning_rate": 5e-05,
"loss": 0.6239,
"step": 2500
},
{
"epoch": 0.85,
"learning_rate": 5e-05,
"loss": 0.6388,
"step": 2510
},
{
"epoch": 0.85,
"learning_rate": 5e-05,
"loss": 0.62,
"step": 2520
},
{
"epoch": 0.85,
"learning_rate": 5e-05,
"loss": 0.6424,
"step": 2530
},
{
"epoch": 0.86,
"learning_rate": 5e-05,
"loss": 0.6368,
"step": 2540
},
{
"epoch": 0.86,
"learning_rate": 5e-05,
"loss": 0.6247,
"step": 2550
},
{
"epoch": 0.86,
"learning_rate": 5e-05,
"loss": 0.624,
"step": 2560
},
{
"epoch": 0.87,
"learning_rate": 5e-05,
"loss": 0.6346,
"step": 2570
},
{
"epoch": 0.87,
"learning_rate": 5e-05,
"loss": 0.6246,
"step": 2580
},
{
"epoch": 0.87,
"learning_rate": 5e-05,
"loss": 0.6177,
"step": 2590
},
{
"epoch": 0.88,
"learning_rate": 5e-05,
"loss": 0.6224,
"step": 2600
},
{
"epoch": 0.88,
"learning_rate": 5e-05,
"loss": 0.6245,
"step": 2610
},
{
"epoch": 0.88,
"learning_rate": 5e-05,
"loss": 0.6248,
"step": 2620
},
{
"epoch": 0.89,
"learning_rate": 5e-05,
"loss": 0.6124,
"step": 2630
},
{
"epoch": 0.89,
"learning_rate": 5e-05,
"loss": 0.632,
"step": 2640
},
{
"epoch": 0.89,
"learning_rate": 5e-05,
"loss": 0.6224,
"step": 2650
},
{
"epoch": 0.9,
"learning_rate": 5e-05,
"loss": 0.6252,
"step": 2660
},
{
"epoch": 0.9,
"learning_rate": 5e-05,
"loss": 0.6227,
"step": 2670
},
{
"epoch": 0.9,
"learning_rate": 5e-05,
"loss": 0.6367,
"step": 2680
},
{
"epoch": 0.91,
"learning_rate": 5e-05,
"loss": 0.641,
"step": 2690
},
{
"epoch": 0.91,
"learning_rate": 5e-05,
"loss": 0.6539,
"step": 2700
},
{
"epoch": 0.91,
"learning_rate": 5e-05,
"loss": 0.6283,
"step": 2710
},
{
"epoch": 0.92,
"learning_rate": 5e-05,
"loss": 0.623,
"step": 2720
},
{
"epoch": 0.92,
"learning_rate": 5e-05,
"loss": 0.6203,
"step": 2730
},
{
"epoch": 0.92,
"learning_rate": 5e-05,
"loss": 0.632,
"step": 2740
},
{
"epoch": 0.93,
"learning_rate": 5e-05,
"loss": 0.6194,
"step": 2750
},
{
"epoch": 0.93,
"learning_rate": 5e-05,
"loss": 0.628,
"step": 2760
},
{
"epoch": 0.93,
"learning_rate": 5e-05,
"loss": 0.6185,
"step": 2770
},
{
"epoch": 0.94,
"learning_rate": 5e-05,
"loss": 0.6254,
"step": 2780
},
{
"epoch": 0.94,
"learning_rate": 5e-05,
"loss": 0.626,
"step": 2790
},
{
"epoch": 0.94,
"learning_rate": 5e-05,
"loss": 0.6322,
"step": 2800
},
{
"epoch": 0.95,
"learning_rate": 5e-05,
"loss": 0.6228,
"step": 2810
},
{
"epoch": 0.95,
"learning_rate": 5e-05,
"loss": 0.6309,
"step": 2820
},
{
"epoch": 0.95,
"learning_rate": 5e-05,
"loss": 0.6261,
"step": 2830
},
{
"epoch": 0.96,
"learning_rate": 5e-05,
"loss": 0.62,
"step": 2840
},
{
"epoch": 0.96,
"learning_rate": 5e-05,
"loss": 0.6271,
"step": 2850
},
{
"epoch": 0.96,
"learning_rate": 5e-05,
"loss": 0.6336,
"step": 2860
},
{
"epoch": 0.97,
"learning_rate": 5e-05,
"loss": 0.6267,
"step": 2870
},
{
"epoch": 0.97,
"learning_rate": 5e-05,
"loss": 0.6182,
"step": 2880
},
{
"epoch": 0.97,
"learning_rate": 5e-05,
"loss": 0.6345,
"step": 2890
},
{
"epoch": 0.98,
"learning_rate": 5e-05,
"loss": 0.6204,
"step": 2900
},
{
"epoch": 0.98,
"learning_rate": 5e-05,
"loss": 0.6172,
"step": 2910
},
{
"epoch": 0.99,
"learning_rate": 5e-05,
"loss": 0.6182,
"step": 2920
},
{
"epoch": 0.99,
"learning_rate": 5e-05,
"loss": 0.6119,
"step": 2930
},
{
"epoch": 0.99,
"learning_rate": 5e-05,
"loss": 0.6178,
"step": 2940
},
{
"epoch": 1.0,
"learning_rate": 5e-05,
"loss": 0.633,
"step": 2950
},
{
"epoch": 1.0,
"learning_rate": 5e-05,
"loss": 0.639,
"step": 2960
},
{
"epoch": 1.0,
"learning_rate": 5e-05,
"loss": 0.6194,
"step": 2970
},
{
"epoch": 1.01,
"learning_rate": 5e-05,
"loss": 0.6148,
"step": 2980
},
{
"epoch": 1.01,
"learning_rate": 5e-05,
"loss": 0.6169,
"step": 2990
},
{
"epoch": 1.01,
"learning_rate": 5e-05,
"loss": 0.6165,
"step": 3000
},
{
"epoch": 1.02,
"learning_rate": 5e-05,
"loss": 0.6319,
"step": 3010
},
{
"epoch": 1.02,
"learning_rate": 5e-05,
"loss": 0.6156,
"step": 3020
},
{
"epoch": 1.02,
"learning_rate": 5e-05,
"loss": 0.6293,
"step": 3030
},
{
"epoch": 1.03,
"learning_rate": 5e-05,
"loss": 0.6196,
"step": 3040
},
{
"epoch": 1.03,
"learning_rate": 5e-05,
"loss": 0.635,
"step": 3050
},
{
"epoch": 1.03,
"learning_rate": 5e-05,
"loss": 0.6436,
"step": 3060
},
{
"epoch": 1.04,
"learning_rate": 5e-05,
"loss": 0.6441,
"step": 3070
},
{
"epoch": 1.04,
"learning_rate": 5e-05,
"loss": 0.628,
"step": 3080
},
{
"epoch": 1.04,
"learning_rate": 5e-05,
"loss": 0.6169,
"step": 3090
},
{
"epoch": 1.05,
"learning_rate": 5e-05,
"loss": 0.6254,
"step": 3100
},
{
"epoch": 1.05,
"learning_rate": 5e-05,
"loss": 0.6246,
"step": 3110
},
{
"epoch": 1.05,
"learning_rate": 5e-05,
"loss": 0.6206,
"step": 3120
},
{
"epoch": 1.06,
"learning_rate": 5e-05,
"loss": 0.6287,
"step": 3130
},
{
"epoch": 1.06,
"learning_rate": 5e-05,
"loss": 0.6256,
"step": 3140
},
{
"epoch": 1.06,
"learning_rate": 5e-05,
"loss": 0.6213,
"step": 3150
},
{
"epoch": 1.07,
"learning_rate": 5e-05,
"loss": 0.6162,
"step": 3160
},
{
"epoch": 1.07,
"learning_rate": 5e-05,
"loss": 0.6265,
"step": 3170
},
{
"epoch": 1.07,
"learning_rate": 5e-05,
"loss": 0.6154,
"step": 3180
},
{
"epoch": 1.08,
"learning_rate": 5e-05,
"loss": 0.6145,
"step": 3190
},
{
"epoch": 1.08,
"learning_rate": 5e-05,
"loss": 0.618,
"step": 3200
},
{
"epoch": 1.08,
"learning_rate": 5e-05,
"loss": 0.6164,
"step": 3210
},
{
"epoch": 1.09,
"learning_rate": 5e-05,
"loss": 0.6169,
"step": 3220
},
{
"epoch": 1.09,
"learning_rate": 5e-05,
"loss": 0.6235,
"step": 3230
},
{
"epoch": 1.09,
"learning_rate": 5e-05,
"loss": 0.6179,
"step": 3240
},
{
"epoch": 1.1,
"learning_rate": 5e-05,
"loss": 0.6156,
"step": 3250
},
{
"epoch": 1.1,
"learning_rate": 5e-05,
"loss": 0.6182,
"step": 3260
},
{
"epoch": 1.1,
"learning_rate": 5e-05,
"loss": 0.6254,
"step": 3270
},
{
"epoch": 1.11,
"learning_rate": 5e-05,
"loss": 0.633,
"step": 3280
},
{
"epoch": 1.11,
"learning_rate": 5e-05,
"loss": 0.6208,
"step": 3290
},
{
"epoch": 1.11,
"learning_rate": 5e-05,
"loss": 0.6172,
"step": 3300
},
{
"epoch": 1.12,
"learning_rate": 5e-05,
"loss": 0.6255,
"step": 3310
},
{
"epoch": 1.12,
"learning_rate": 5e-05,
"loss": 0.6178,
"step": 3320
},
{
"epoch": 1.12,
"learning_rate": 5e-05,
"loss": 0.6154,
"step": 3330
},
{
"epoch": 1.13,
"learning_rate": 5e-05,
"loss": 0.6198,
"step": 3340
},
{
"epoch": 1.13,
"learning_rate": 5e-05,
"loss": 0.6141,
"step": 3350
},
{
"epoch": 1.13,
"learning_rate": 5e-05,
"loss": 0.6173,
"step": 3360
},
{
"epoch": 1.14,
"learning_rate": 5e-05,
"loss": 0.625,
"step": 3370
},
{
"epoch": 1.14,
"learning_rate": 5e-05,
"loss": 0.6285,
"step": 3380
},
{
"epoch": 1.14,
"learning_rate": 5e-05,
"loss": 0.6147,
"step": 3390
},
{
"epoch": 1.15,
"learning_rate": 5e-05,
"loss": 0.6119,
"step": 3400
},
{
"epoch": 1.15,
"learning_rate": 5e-05,
"loss": 0.6215,
"step": 3410
},
{
"epoch": 1.15,
"learning_rate": 5e-05,
"loss": 0.6203,
"step": 3420
},
{
"epoch": 1.16,
"learning_rate": 5e-05,
"loss": 0.6186,
"step": 3430
},
{
"epoch": 1.16,
"learning_rate": 5e-05,
"loss": 0.6221,
"step": 3440
},
{
"epoch": 1.16,
"learning_rate": 5e-05,
"loss": 0.6183,
"step": 3450
},
{
"epoch": 1.17,
"learning_rate": 5e-05,
"loss": 0.6129,
"step": 3460
},
{
"epoch": 1.17,
"learning_rate": 5e-05,
"loss": 0.6267,
"step": 3470
},
{
"epoch": 1.17,
"learning_rate": 5e-05,
"loss": 0.625,
"step": 3480
},
{
"epoch": 1.18,
"learning_rate": 5e-05,
"loss": 0.6183,
"step": 3490
},
{
"epoch": 1.18,
"learning_rate": 5e-05,
"loss": 0.6201,
"step": 3500
},
{
"epoch": 1.18,
"learning_rate": 5e-05,
"loss": 0.6242,
"step": 3510
},
{
"epoch": 1.19,
"learning_rate": 5e-05,
"loss": 0.6144,
"step": 3520
},
{
"epoch": 1.19,
"learning_rate": 5e-05,
"loss": 0.6136,
"step": 3530
},
{
"epoch": 1.19,
"learning_rate": 5e-05,
"loss": 0.6217,
"step": 3540
},
{
"epoch": 1.2,
"learning_rate": 5e-05,
"loss": 0.6214,
"step": 3550
},
{
"epoch": 1.2,
"learning_rate": 5e-05,
"loss": 0.6186,
"step": 3560
},
{
"epoch": 1.2,
"learning_rate": 5e-05,
"loss": 0.6227,
"step": 3570
},
{
"epoch": 1.21,
"learning_rate": 5e-05,
"loss": 0.6185,
"step": 3580
},
{
"epoch": 1.21,
"learning_rate": 5e-05,
"loss": 0.633,
"step": 3590
},
{
"epoch": 1.21,
"learning_rate": 5e-05,
"loss": 0.6179,
"step": 3600
},
{
"epoch": 1.22,
"learning_rate": 5e-05,
"loss": 0.6236,
"step": 3610
},
{
"epoch": 1.22,
"learning_rate": 5e-05,
"loss": 0.6251,
"step": 3620
},
{
"epoch": 1.22,
"learning_rate": 5e-05,
"loss": 0.6214,
"step": 3630
},
{
"epoch": 1.23,
"learning_rate": 5e-05,
"loss": 0.6091,
"step": 3640
},
{
"epoch": 1.23,
"learning_rate": 5e-05,
"loss": 0.6235,
"step": 3650
},
{
"epoch": 1.23,
"learning_rate": 5e-05,
"loss": 0.6367,
"step": 3660
},
{
"epoch": 1.24,
"learning_rate": 5e-05,
"loss": 0.6197,
"step": 3670
},
{
"epoch": 1.24,
"learning_rate": 5e-05,
"loss": 0.6178,
"step": 3680
},
{
"epoch": 1.24,
"learning_rate": 5e-05,
"loss": 0.6089,
"step": 3690
},
{
"epoch": 1.25,
"learning_rate": 5e-05,
"loss": 0.6199,
"step": 3700
},
{
"epoch": 1.25,
"learning_rate": 5e-05,
"loss": 0.6111,
"step": 3710
},
{
"epoch": 1.25,
"learning_rate": 5e-05,
"loss": 0.6084,
"step": 3720
},
{
"epoch": 1.26,
"learning_rate": 5e-05,
"loss": 0.6159,
"step": 3730
},
{
"epoch": 1.26,
"learning_rate": 5e-05,
"loss": 0.6213,
"step": 3740
},
{
"epoch": 1.27,
"learning_rate": 5e-05,
"loss": 0.6138,
"step": 3750
},
{
"epoch": 1.27,
"learning_rate": 5e-05,
"loss": 0.6159,
"step": 3760
},
{
"epoch": 1.27,
"learning_rate": 5e-05,
"loss": 0.6145,
"step": 3770
},
{
"epoch": 1.28,
"learning_rate": 5e-05,
"loss": 0.6195,
"step": 3780
},
{
"epoch": 1.28,
"learning_rate": 5e-05,
"loss": 0.6179,
"step": 3790
},
{
"epoch": 1.28,
"learning_rate": 5e-05,
"loss": 0.611,
"step": 3800
},
{
"epoch": 1.29,
"learning_rate": 5e-05,
"loss": 0.6145,
"step": 3810
},
{
"epoch": 1.29,
"learning_rate": 5e-05,
"loss": 0.611,
"step": 3820
},
{
"epoch": 1.29,
"learning_rate": 5e-05,
"loss": 0.6113,
"step": 3830
},
{
"epoch": 1.3,
"learning_rate": 5e-05,
"loss": 0.6107,
"step": 3840
},
{
"epoch": 1.3,
"learning_rate": 5e-05,
"loss": 0.6096,
"step": 3850
},
{
"epoch": 1.3,
"learning_rate": 5e-05,
"loss": 0.6218,
"step": 3860
},
{
"epoch": 1.31,
"learning_rate": 5e-05,
"loss": 0.6192,
"step": 3870
},
{
"epoch": 1.31,
"learning_rate": 5e-05,
"loss": 0.6057,
"step": 3880
},
{
"epoch": 1.31,
"learning_rate": 5e-05,
"loss": 0.615,
"step": 3890
},
{
"epoch": 1.32,
"learning_rate": 5e-05,
"loss": 0.6248,
"step": 3900
},
{
"epoch": 1.32,
"learning_rate": 5e-05,
"loss": 0.6166,
"step": 3910
},
{
"epoch": 1.32,
"learning_rate": 5e-05,
"loss": 0.618,
"step": 3920
},
{
"epoch": 1.33,
"learning_rate": 5e-05,
"loss": 0.6122,
"step": 3930
},
{
"epoch": 1.33,
"learning_rate": 5e-05,
"loss": 0.6261,
"step": 3940
},
{
"epoch": 1.33,
"learning_rate": 5e-05,
"loss": 0.6094,
"step": 3950
},
{
"epoch": 1.34,
"learning_rate": 5e-05,
"loss": 0.6158,
"step": 3960
},
{
"epoch": 1.34,
"learning_rate": 5e-05,
"loss": 0.616,
"step": 3970
},
{
"epoch": 1.34,
"learning_rate": 5e-05,
"loss": 0.6114,
"step": 3980
},
{
"epoch": 1.35,
"learning_rate": 5e-05,
"loss": 0.6038,
"step": 3990
},
{
"epoch": 1.35,
"learning_rate": 5e-05,
"loss": 0.606,
"step": 4000
},
{
"epoch": 1.35,
"learning_rate": 5e-05,
"loss": 0.6151,
"step": 4010
},
{
"epoch": 1.36,
"learning_rate": 5e-05,
"loss": 0.6044,
"step": 4020
},
{
"epoch": 1.36,
"learning_rate": 5e-05,
"loss": 0.6222,
"step": 4030
},
{
"epoch": 1.36,
"learning_rate": 5e-05,
"loss": 0.6114,
"step": 4040
},
{
"epoch": 1.37,
"learning_rate": 5e-05,
"loss": 0.6043,
"step": 4050
},
{
"epoch": 1.37,
"learning_rate": 5e-05,
"loss": 0.6113,
"step": 4060
},
{
"epoch": 1.37,
"learning_rate": 5e-05,
"loss": 0.6134,
"step": 4070
},
{
"epoch": 1.38,
"learning_rate": 5e-05,
"loss": 0.6255,
"step": 4080
},
{
"epoch": 1.38,
"learning_rate": 5e-05,
"loss": 0.6099,
"step": 4090
},
{
"epoch": 1.38,
"learning_rate": 5e-05,
"loss": 0.6168,
"step": 4100
},
{
"epoch": 1.39,
"learning_rate": 5e-05,
"loss": 0.6127,
"step": 4110
},
{
"epoch": 1.39,
"learning_rate": 5e-05,
"loss": 0.611,
"step": 4120
},
{
"epoch": 1.39,
"learning_rate": 5e-05,
"loss": 0.601,
"step": 4130
},
{
"epoch": 1.4,
"learning_rate": 5e-05,
"loss": 0.611,
"step": 4140
},
{
"epoch": 1.4,
"learning_rate": 5e-05,
"loss": 0.6081,
"step": 4150
},
{
"epoch": 1.4,
"learning_rate": 5e-05,
"loss": 0.6159,
"step": 4160
},
{
"epoch": 1.41,
"learning_rate": 5e-05,
"loss": 0.6101,
"step": 4170
},
{
"epoch": 1.41,
"learning_rate": 5e-05,
"loss": 0.6038,
"step": 4180
},
{
"epoch": 1.41,
"learning_rate": 5e-05,
"loss": 0.6061,
"step": 4190
},
{
"epoch": 1.42,
"learning_rate": 5e-05,
"loss": 0.6094,
"step": 4200
},
{
"epoch": 1.42,
"learning_rate": 5e-05,
"loss": 0.6031,
"step": 4210
},
{
"epoch": 1.42,
"learning_rate": 5e-05,
"loss": 0.6237,
"step": 4220
},
{
"epoch": 1.43,
"learning_rate": 5e-05,
"loss": 0.6298,
"step": 4230
},
{
"epoch": 1.43,
"learning_rate": 5e-05,
"loss": 0.6109,
"step": 4240
},
{
"epoch": 1.43,
"learning_rate": 5e-05,
"loss": 0.6045,
"step": 4250
},
{
"epoch": 1.44,
"learning_rate": 5e-05,
"loss": 0.6121,
"step": 4260
},
{
"epoch": 1.44,
"learning_rate": 5e-05,
"loss": 0.6138,
"step": 4270
},
{
"epoch": 1.44,
"learning_rate": 5e-05,
"loss": 0.6099,
"step": 4280
},
{
"epoch": 1.45,
"learning_rate": 5e-05,
"loss": 0.6165,
"step": 4290
},
{
"epoch": 1.45,
"learning_rate": 5e-05,
"loss": 0.6076,
"step": 4300
},
{
"epoch": 1.45,
"learning_rate": 5e-05,
"loss": 0.617,
"step": 4310
},
{
"epoch": 1.46,
"learning_rate": 5e-05,
"loss": 0.6109,
"step": 4320
},
{
"epoch": 1.46,
"learning_rate": 5e-05,
"loss": 0.6132,
"step": 4330
},
{
"epoch": 1.46,
"learning_rate": 5e-05,
"loss": 0.6126,
"step": 4340
},
{
"epoch": 1.47,
"learning_rate": 5e-05,
"loss": 0.6268,
"step": 4350
},
{
"epoch": 1.47,
"learning_rate": 5e-05,
"loss": 0.6063,
"step": 4360
},
{
"epoch": 1.47,
"learning_rate": 5e-05,
"loss": 0.6052,
"step": 4370
},
{
"epoch": 1.48,
"learning_rate": 5e-05,
"loss": 0.6072,
"step": 4380
},
{
"epoch": 1.48,
"learning_rate": 5e-05,
"loss": 0.6044,
"step": 4390
},
{
"epoch": 1.48,
"learning_rate": 5e-05,
"loss": 0.6137,
"step": 4400
},
{
"epoch": 1.49,
"learning_rate": 5e-05,
"loss": 0.5983,
"step": 4410
},
{
"epoch": 1.49,
"learning_rate": 5e-05,
"loss": 0.6113,
"step": 4420
},
{
"epoch": 1.49,
"learning_rate": 5e-05,
"loss": 0.6158,
"step": 4430
},
{
"epoch": 1.5,
"learning_rate": 5e-05,
"loss": 0.6079,
"step": 4440
},
{
"epoch": 1.5,
"learning_rate": 5e-05,
"loss": 0.6128,
"step": 4450
},
{
"epoch": 1.5,
"learning_rate": 5e-05,
"loss": 0.6222,
"step": 4460
},
{
"epoch": 1.51,
"learning_rate": 5e-05,
"loss": 0.6085,
"step": 4470
},
{
"epoch": 1.51,
"learning_rate": 5e-05,
"loss": 0.6161,
"step": 4480
},
{
"epoch": 1.51,
"learning_rate": 5e-05,
"loss": 0.6099,
"step": 4490
},
{
"epoch": 1.52,
"learning_rate": 5e-05,
"loss": 0.613,
"step": 4500
},
{
"epoch": 1.52,
"learning_rate": 5e-05,
"loss": 0.6121,
"step": 4510
},
{
"epoch": 1.52,
"learning_rate": 5e-05,
"loss": 0.6106,
"step": 4520
},
{
"epoch": 1.53,
"learning_rate": 5e-05,
"loss": 0.6151,
"step": 4530
},
{
"epoch": 1.53,
"learning_rate": 5e-05,
"loss": 0.6173,
"step": 4540
},
{
"epoch": 1.53,
"learning_rate": 5e-05,
"loss": 0.6166,
"step": 4550
},
{
"epoch": 1.54,
"learning_rate": 5e-05,
"loss": 0.6099,
"step": 4560
},
{
"epoch": 1.54,
"learning_rate": 5e-05,
"loss": 0.6216,
"step": 4570
},
{
"epoch": 1.55,
"learning_rate": 5e-05,
"loss": 0.6115,
"step": 4580
},
{
"epoch": 1.55,
"learning_rate": 5e-05,
"loss": 0.608,
"step": 4590
},
{
"epoch": 1.55,
"learning_rate": 5e-05,
"loss": 0.6124,
"step": 4600
},
{
"epoch": 1.56,
"learning_rate": 5e-05,
"loss": 0.6071,
"step": 4610
},
{
"epoch": 1.56,
"learning_rate": 5e-05,
"loss": 0.6052,
"step": 4620
},
{
"epoch": 1.56,
"learning_rate": 5e-05,
"loss": 0.6158,
"step": 4630
},
{
"epoch": 1.57,
"learning_rate": 5e-05,
"loss": 0.6163,
"step": 4640
},
{
"epoch": 1.57,
"learning_rate": 5e-05,
"loss": 0.6088,
"step": 4650
},
{
"epoch": 1.57,
"learning_rate": 5e-05,
"loss": 0.6129,
"step": 4660
},
{
"epoch": 1.58,
"learning_rate": 5e-05,
"loss": 0.6284,
"step": 4670
},
{
"epoch": 1.58,
"learning_rate": 5e-05,
"loss": 0.6104,
"step": 4680
},
{
"epoch": 1.58,
"learning_rate": 5e-05,
"loss": 0.6017,
"step": 4690
},
{
"epoch": 1.59,
"learning_rate": 5e-05,
"loss": 0.6118,
"step": 4700
},
{
"epoch": 1.59,
"learning_rate": 5e-05,
"loss": 0.6089,
"step": 4710
},
{
"epoch": 1.59,
"learning_rate": 5e-05,
"loss": 0.6158,
"step": 4720
},
{
"epoch": 1.6,
"learning_rate": 5e-05,
"loss": 0.6124,
"step": 4730
},
{
"epoch": 1.6,
"learning_rate": 5e-05,
"loss": 0.6044,
"step": 4740
},
{
"epoch": 1.6,
"learning_rate": 5e-05,
"loss": 0.604,
"step": 4750
},
{
"epoch": 1.61,
"learning_rate": 5e-05,
"loss": 0.6189,
"step": 4760
},
{
"epoch": 1.61,
"learning_rate": 5e-05,
"loss": 0.614,
"step": 4770
},
{
"epoch": 1.61,
"learning_rate": 5e-05,
"loss": 0.6298,
"step": 4780
},
{
"epoch": 1.62,
"learning_rate": 5e-05,
"loss": 0.6215,
"step": 4790
},
{
"epoch": 1.62,
"learning_rate": 5e-05,
"loss": 0.6018,
"step": 4800
},
{
"epoch": 1.62,
"learning_rate": 5e-05,
"loss": 0.6107,
"step": 4810
},
{
"epoch": 1.63,
"learning_rate": 5e-05,
"loss": 0.6027,
"step": 4820
},
{
"epoch": 1.63,
"learning_rate": 5e-05,
"loss": 0.615,
"step": 4830
},
{
"epoch": 1.63,
"learning_rate": 5e-05,
"loss": 0.605,
"step": 4840
},
{
"epoch": 1.64,
"learning_rate": 5e-05,
"loss": 0.6094,
"step": 4850
},
{
"epoch": 1.64,
"learning_rate": 5e-05,
"loss": 0.6073,
"step": 4860
},
{
"epoch": 1.64,
"learning_rate": 5e-05,
"loss": 0.6069,
"step": 4870
},
{
"epoch": 1.65,
"learning_rate": 5e-05,
"loss": 0.6027,
"step": 4880
},
{
"epoch": 1.65,
"learning_rate": 5e-05,
"loss": 0.6127,
"step": 4890
},
{
"epoch": 1.65,
"learning_rate": 5e-05,
"loss": 0.6178,
"step": 4900
},
{
"epoch": 1.66,
"learning_rate": 5e-05,
"loss": 0.5996,
"step": 4910
},
{
"epoch": 1.66,
"learning_rate": 5e-05,
"loss": 0.6091,
"step": 4920
},
{
"epoch": 1.66,
"learning_rate": 5e-05,
"loss": 0.6051,
"step": 4930
},
{
"epoch": 1.67,
"learning_rate": 5e-05,
"loss": 0.6187,
"step": 4940
},
{
"epoch": 1.67,
"learning_rate": 5e-05,
"loss": 0.6234,
"step": 4950
},
{
"epoch": 1.67,
"learning_rate": 5e-05,
"loss": 0.6133,
"step": 4960
},
{
"epoch": 1.68,
"learning_rate": 5e-05,
"loss": 0.6057,
"step": 4970
},
{
"epoch": 1.68,
"learning_rate": 5e-05,
"loss": 0.6117,
"step": 4980
},
{
"epoch": 1.68,
"learning_rate": 5e-05,
"loss": 0.6151,
"step": 4990
},
{
"epoch": 1.69,
"learning_rate": 5e-05,
"loss": 0.6083,
"step": 5000
}
],
"logging_steps": 10,
"max_steps": 8892,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"save_steps": 1000,
"total_flos": 0.0,
"train_batch_size": 8,
"trial_name": null,
"trial_params": null
}